Propellerhead Software

Go Back   Propellerhead Forum > General Forum (read only)

 
 
Thread Tools Display Modes
  #1  
Old 2012-01-06, 02:03
auschris auschris is offline
 
Join Date: Nov 2007
Posts: 101
Gain or Normalize audio tracks for mixing

Im a novice when it comes to mixing.
I have been reading about the importance of gain staging before starting to mix tracks ie setting all tracks to 0db.
On a recent mix i used the gain knob on the input section of the reason 6 mixer to set all my audio tracks to around 0db before starting to apply pan / fader moves etc, later i thought it would have been easier to just normalize each audio file as this would have taken some of the guess work out of manually setting each channel to 0db.

My question is: is normalizing audio tracks an appropriate way to set gain structure post recording?

Does anyone do this?

Chris
  #2  
Old 2012-01-06, 03:51
3rdFloorSound's Avatar
3rdFloorSound 3rdFloorSound is offline
 
Join Date: Aug 2010
Posts: 10,766
I do something sort of like that. I'll slice sections that are too quiet around a part that ought to be at peak and normalize the section, but you have to mind the slice point to make sure there aren't large volume leaps next to the other clips.
__________________
  #3  
Old 2012-01-06, 03:56
Anomecron's Avatar
Anomecron Anomecron is offline
 
Join Date: Jan 2004
Posts: 2,904
Quote:
Originally Posted by auschris View Post
Im a novice when it comes to mixing.
I have been reading about the importance of gain staging before starting to mix tracks ie setting all tracks to 0db.
On a recent mix i used the gain knob on the input section of the reason 6 mixer to set all my audio tracks to around 0db before starting to apply pan / fader moves etc, later i thought it would have been easier to just normalize each audio file as this would have taken some of the guess work out of manually setting each channel to 0db.

My question is: is normalizing audio tracks an appropriate way to set gain structure post recording?

Does anyone do this?

Chris

When sampling I tend to normalise but be weary that any noise inherent in the source file will be amplified also - so ideally you should record the source as cleanly as possible and as close to 0DB as you can.
I am quite often prone to checking files in an audio editor first for DC offset or other noise problems and then bringing them back into Reason also but your mileage may vary.
  #4  
Old 2012-01-06, 04:41
selig's Avatar
selig selig is offline
 
Join Date: Jun 2003
Posts: 10,517
Quote:
Originally Posted by Anomecron View Post
When sampling I tend to normalise but be weary that any noise inherent in the source file will be amplified also - so ideally you should record the source as cleanly as possible and as close to 0DB as you can.
I am quite often prone to checking files in an audio editor first for DC offset or other noise problems and then bringing them back into Reason also but your mileage may vary.
Since, as you say, when you raise a track the noise will be amplified also, there's nothing to worry about - it will sound EXACTLY the same after you raise it! It's not like the noise gets any louder than the rest of the signal when you do this!

Having said that, I'm careful to suggest that folks use PEAK metering and set all levels to peak around -12 dBFS, and not to normalize to 0 dB. The channel meters are VU, and what you're calling "0 dB" is actually -12 dBFS. BUT since this is VU, the peak for each channel will often be much different (and always HIGHER) than the VU, which causes some tracks to clip the output sooner than others.

To set levels to -12 dBFS Peaks, use the main meter, solo the track in question, and for instrument tracks I use the instrument's master level. For audio tracks, you should record them at the same level, peaking around -12 dBFS. If you import tracks from somewhere else, you can use the clip gain in the sequencer to bring the level to around -12 dBFS on the peak meters. :-)
__________________
Giles Reaves, aka 'selig'
Audio Illusionist, Musical Technologist
Selig Audio, LLC
USA
  #5  
Old 2012-01-06, 08:27
juno106user's Avatar
juno106user juno106user is offline
 
Join Date: Jun 2007
Posts: 490
I normalize files about 100 times per day.. No joke! I realize that it is less desirable than getting the perfect input levels when recording but when you are constantly churning out material, normalizing sections can be an easy way to quickly edit your way through your material.
When recording dynamic sources (certain people's voices for example), normalization can be redundant if a "T" or a "P" sound's attack is much louder than the body of the word. It will only normalize in relation to the peak level of the "T&P" sounds thus the overall gain can be insignificant as the body of the word will still remain lower than desired.
I work in a decent studio with a pretty low noise floor which then allows me to punch up the overall dynamics using some carefully placed compression from area to area. So, although I prefer to normalize sections because as selig indicated the overall ratio between signal to noise remains the same, my recordings have that low noise floor so I can compress when needed and the increase in low level noise remains relatively insignificant.
I will insert a compressor on a channel if I think it is necessary but the techniques discussed above are destructive so the original files are replaced in my session.
I almost never use gain but I almost always do volume automation on tracks. Drums, synths, vocals, sound effects, whatever.. I think that static volume levels create less interesting material not to mention the practical reasons for automating volume.
  #6  
Old 2012-01-06, 22:43
selig's Avatar
selig selig is offline
 
Join Date: Jun 2003
Posts: 10,517
Quote:
Originally Posted by juno106user View Post
I normalize files about 100 times per day.. No joke! I realize that it is less desirable than getting the perfect input levels when recording but when you are constantly churning out material, normalizing sections can be an easy way to quickly edit your way through your material.
When recording dynamic sources (certain people's voices for example), normalization can be redundant if a "T" or a "P" sound's attack is much louder than the body of the word. It will only normalize in relation to the peak level of the "T&P" sounds thus the overall gain can be insignificant as the body of the word will still remain lower than desired.
I work in a decent studio with a pretty low noise floor which then allows me to punch up the overall dynamics using some carefully placed compression from area to area. So, although I prefer to normalize sections because as selig indicated the overall ratio between signal to noise remains the same, my recordings have that low noise floor so I can compress when needed and the increase in low level noise remains relatively insignificant.
I will insert a compressor on a channel if I think it is necessary but the techniques discussed above are destructive so the original files are replaced in my session.
I almost never use gain but I almost always do volume automation on tracks. Drums, synths, vocals, sound effects, whatever.. I think that static volume levels create less interesting material not to mention the practical reasons for automating volume.
Are you saying that all your audio tracks are peaking @ 0dBFS (what about instrument tracks)? I find that's the quickest way to clip a mix, and if you were to lower each track by about 12 dB you'd be good - OR just start with them at that level and never worry again! ;-)

I find it VERY helpful when working quickly to have ALL audio at the same level (audio and inst tracks), which means all my insert FX will see the same basic level and make it easy to not have surprises at any stage of production. All in the name of speed and predictability! :-)
__________________
Giles Reaves, aka 'selig'
Audio Illusionist, Musical Technologist
Selig Audio, LLC
USA
  #7  
Old 2012-01-07, 00:33
juno106user's Avatar
juno106user juno106user is offline
 
Join Date: Jun 2007
Posts: 490
Exactly.. As a rule, I always start my tracks out at -6 and if I need to increase an area, I have headroom, including on the master. It's just habit and it works for me. It's not that uncommon to grab everything at once and pull it all down further if I've been pushing a lot. Which reminds me of a studio trick that I heard Dr. Dre uses (not actually sure if it's true but..) I was told that he always pushes when he mixes so at many times during the mixing process, he'll keep dropping everything across the board by -6dBFS (or some common increment) and keep pushing and pushing! Kind of a cool idea.

Last edited by juno106user; 2012-01-07 at 00:38.
  #8  
Old 2012-01-09, 05:59
Anomecron's Avatar
Anomecron Anomecron is offline
 
Join Date: Jan 2004
Posts: 2,904
Quote:
Originally Posted by selig View Post
Since, as you say, when you raise a track the noise will be amplified also, there's nothing to worry about - it will sound EXACTLY the same after you raise it! It's not like the noise gets any louder than the rest of the signal when you do this!

Having said that, I'm careful to suggest that folks use PEAK metering and set all levels to peak around -12 dBFS, and not to normalize to 0 dB. The channel meters are VU, and what you're calling "0 dB" is actually -12 dBFS. BUT since this is VU, the peak for each channel will often be much different (and always HIGHER) than the VU, which causes some tracks to clip the output sooner than others.

To set levels to -12 dBFS Peaks, use the main meter, solo the track in question, and for instrument tracks I use the instrument's master level. For audio tracks, you should record them at the same level, peaking around -12 dBFS. If you import tracks from somewhere else, you can use the clip gain in the sequencer to bring the level to around -12 dBFS on the peak meters. :-)


True points and your first paragraph made me giggle at myself for in fact stating the obvious (or at least having it pointed out to me).
I was referring more to the old sampling adage of junk in junk out - in the sense that if the source audio isn't up to snuff normalising will only enhance the problems inherent in the source file itself (which is stating the obvious once again) so I will just shut up and quit whilst I'm ahead I think.
Interesting points on setting levels to -12dBFS and it's always great to apply new techniques to the way I work - which I have just done on the Broadfield remix competition and really pleased with the end results thus far I must say.
You are a bottomless pit of information and I am not that up on a lot of technical jargon.
I'm largely self taught and go purely on ears and intuition when mixing which has meant a somewhat anarchic approach to production which has become more controlled and tempered over the years.
I know things work for specific reasons through function and experimenting but know very little about the technical engineering aspects of what is happening so you always provide a) great intuitive explanations and b) great insider insight and information.
Well enough prattling from me.
Thanks for the input once again.
  #9  
Old 2012-01-09, 17:04
illCarl's Avatar
illCarl illCarl is offline
 
Join Date: Oct 2009
Posts: 691
Quote:
Originally Posted by selig View Post
Since, as you say, when you raise a track the noise will be amplified also, there's nothing to worry about - it will sound EXACTLY the same after you raise it! It's not like the noise gets any louder than the rest of the signal when you do this!
Hi

Was just wondering.. but that sounds kinda wrong to me.. My understanding of normalising is that every bit of the signal is raised to the max, so you basically loose all dynamics. Since background noise is quite low volume, won't it sound much louder after normalising?

I might be wrong though I never use it
  #10  
Old 2012-01-09, 17:18
selig's Avatar
selig selig is offline
 
Join Date: Jun 2003
Posts: 10,517
Quote:
Originally Posted by illCarl View Post
Hi

Was just wondering.. but that sounds kinda wrong to me.. My understanding of normalising is that every bit of the signal is raised to the max, so you basically loose all dynamics. Since background noise is quite low volume, won't it sound much louder after normalising?

I might be wrong though I never use it
Normalizing is just adding gain, nothing more, nothing less. The only additional step in the processes is that before any gain is added, the existing file is analyzed for it's highest peak level in the entire file, and then the difference between that peak level and clipping is calculated and added to the file.

For example, if the highest peak was @ -6 dBFS, then normalizing will add 6 dB gain to the file. Since 6 dB is added to ALL audio in the file equally, there's no way any part of it can become 'louder'.

But to answer your last question, yes the background noise will be louder after normalizing, but so will EVERYTHING ELSE! In the last example, the background noise will be 6 dB louder, but so will even the loudest signal, leaving the dynamic range of the audio un-affected.

Normalization isn't compression, it's just gain. Others have said not to normalize for "sonic" reasons, but that's like saying "Don't ever change the gain of any audio"! If a digital audio system can't handle a simple gain change without messing with space, depth, and dimension, then I'd say far away from it! ;-)

My reason for not normalizing for audio production has to do with wasting time and steps - it's usually un-necessary and typically ends up getting "un-done" in the mix process where you have to lower all your levels to prevent clipping the outputs. :-)
__________________
Giles Reaves, aka 'selig'
Audio Illusionist, Musical Technologist
Selig Audio, LLC
USA
 

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Fastest way to export all tracks (audio and midi) to stems? Upright Phead User Forum (read only) 2 2011-04-28 06:49
Didn't realise how useful unlimited audio tracks were......... Puzzlefactory Phead User Forum (read only) 3 2010-10-07 16:56
Record: Severe performance issues with 50+ audio tracks sonicnurse Phead User Forum (read only) 67 2010-05-07 21:16
Audio tracks and Mix Tracks: Need help Plasma Phead User Forum (read only) 9 2009-10-12 00:20


All times are GMT +2. The time now is 03:27.